Table 3.

Overview on the quality of taxonomy induction

Hearst (1992)	Precision > 90%, Recall << 10%
Caraballo (1999)	33% Precision (strict), 60% Precision
Sanderson and Croft (1999)	48% Precision (baseline 28%)
Cimiano et al. (2005)	F=0.33 (Finance), F=0.41 (Tourism)
Snow et al. (2004)	Maximal F-measure 14.2–35.9%
Snow et al. (2006)	58% Precision, 20% Recall
Ryu and Choi (2006)	All Recall and Precision below 50%

The F-measure is usually <50%. In information retrieval, quality is often measured as F-measure (F), the harmonic mean of precision and recall (see ‘Methods’ section).